Paper Info
Title
How important are updating schemes in multi-agent systems? An illustration on a multi-turmite model
Abstract
It is to date an open question to know how the updating methods affect the evolution of a multi-agent system. This question has been tackled for various complex systems such as cellular automata, Boolean networks, neural networks but little is known for multi-agent systems, especially for the models with a complex behaviour which emerges from simple local rules. This paper focuses on a multi-turmite model, namely the multiple Langton's ants model. All the agents are updated simultaneously and the variation of the updating scheme consists only in choosing different strategies for solving the conflicts produced when two or more agents want to go on the same location. We show that for the same formulation of the agents' behaviour, and the same initial conditions, the use of different updating schemes may lead to qualitatively different evolutions of the system. As a positive spin-off of this study, we exhibit new phenomena of the multi-turmite model such as deadlocks or gliders.
Year
DOI
Venue
2010
10.5555/1838206.1838282
AAMAS
Keywords
Field
DocType
complex behaviour,different strategy,different evolution,ants model,cellular automaton,multi-turmite model,multi-agent system,open question,boolean network,various complex system,neural network,multi agent system,complex system,cellular automata,initial condition,complex systems
Complex system,Cellular automaton,Know-how,Computer science,Deadlock,Multi-agent system,Turmite,Artificial intelligence,Artificial neural network,Machine learning
Conference
Citations 
PageRank 
References 
6
0.58
8
Authors
2
Name
Order
Citations
PageRank
Nazim Fatès121225.31
Vincent Chevrier215724.47